November 21, 2006 at 16:43Recommendation systems as the apotheosis of the consumer society
Social recommendation systems are based on special algorithms that analyze user preferences, recognize common patterns, and try to predict which product may still be of interest to a specific user.
Modern recommendation systems are able to make the most accurate psychological portrait of a person on the basis of minimal initial data: it is enough to know a few of his favorite films and books. Sometimes a single name is enough (if it is not very common). After that, you can confidently predict what products this person wants to buy, and indeed - to which class of consumers he belongs from the point of view of psychography .
In a developed consumer society, any recommendation system that can increase sales by at least 1% will bring its creator a fortune. For the creation of such a system right now you can get a million dollar prize . It is not surprising that recently in the wake of the boom of Web 2.0 on the American Internet, one after another , new recommendation engines have appeared in various fields: for music, books, websites, TV shows, other people, etc.
Previously, recommendation systems worked only in specific product groups. For example, knowing about the purchase of some books, they could recommend other books. Now, such systems are capable of “transforming" preferences from one commodity reality to another. Knowing your favorite movie - pick up clothes. According to the information about the car brand, choose furniture for the home. The system is able to select the goods according to the psychological profile of the person.
For example, in the What to Rent system, you must pass an unusual psychological test of twenty questions, including telling about your first love. After that, you need to indicate what mood you are in right now. The system analyzes all the information and displays the name of the movie that suits you best. If you have already watched it, another one appears.
The system was created by two technical students and at the same time video fans. At first glance, they can determine the favorite movie of almost every person on the street, and so they decided to create a computer system that does the same.
Recommendation systems like What to Rent appear on the Internet like mushrooms after rain: MyStrands , StumbleUpon , Pandora.com , CleverSet , ChoiceStream (this engine is used in iTunes and DirecTV) and many others.
Of course, the idea is by no means new. Large retail chains, representatives of the advertising industry have long been solving this problem. Despite some success (the emergence of smart online advertising), research in this direction is ongoing.
In a world where consumption and advertising reign, every person has a consumer pattern. Each of us gives preference to certain brands, brands, certain products. Even if a person specifically ignores brands, this is also a certain pattern that clearly speaks of exactly what and how to sell such a person. Modern recommendation systems aim to detect such patterns.
This is a good, very promising business. Especially now, when every person has an archive of digital data, according to which his consumer portrait can be accurately built in a split second. An archive of digital music in an MP3 player, a collection of digital films, a history of surfing, tags and bookmarks on Web 2.0 sites - and you're done. After analyzing this information, the person fell on the hook of a marketer. Now he will receive commercial offers, which he will not be able to refuse. The point is small - to create a universal system that can automate this process.
Hundreds of different companies are working on this challenge. Even former Kiev resident Max Levchin (founder of Paypal ) created a new startup Slidein this area. He says, by the way, that Paypal is also equipped with something like a recommendation system: they analyze the profile of each user and calculate what the probability of fraud on his part.
Perhaps the next giant of Internet search, the so-called “smart Google” , will be created on the basis of these new wave companies.
Modern recommendation systems are able to make the most accurate psychological portrait of a person on the basis of minimal initial data: it is enough to know a few of his favorite films and books. Sometimes a single name is enough (if it is not very common). After that, you can confidently predict what products this person wants to buy, and indeed - to which class of consumers he belongs from the point of view of psychography .
In a developed consumer society, any recommendation system that can increase sales by at least 1% will bring its creator a fortune. For the creation of such a system right now you can get a million dollar prize . It is not surprising that recently in the wake of the boom of Web 2.0 on the American Internet, one after another , new recommendation engines have appeared in various fields: for music, books, websites, TV shows, other people, etc.
Previously, recommendation systems worked only in specific product groups. For example, knowing about the purchase of some books, they could recommend other books. Now, such systems are capable of “transforming" preferences from one commodity reality to another. Knowing your favorite movie - pick up clothes. According to the information about the car brand, choose furniture for the home. The system is able to select the goods according to the psychological profile of the person.
For example, in the What to Rent system, you must pass an unusual psychological test of twenty questions, including telling about your first love. After that, you need to indicate what mood you are in right now. The system analyzes all the information and displays the name of the movie that suits you best. If you have already watched it, another one appears.
The system was created by two technical students and at the same time video fans. At first glance, they can determine the favorite movie of almost every person on the street, and so they decided to create a computer system that does the same.
Recommendation systems like What to Rent appear on the Internet like mushrooms after rain: MyStrands , StumbleUpon , Pandora.com , CleverSet , ChoiceStream (this engine is used in iTunes and DirecTV) and many others.
Of course, the idea is by no means new. Large retail chains, representatives of the advertising industry have long been solving this problem. Despite some success (the emergence of smart online advertising), research in this direction is ongoing.
In a world where consumption and advertising reign, every person has a consumer pattern. Each of us gives preference to certain brands, brands, certain products. Even if a person specifically ignores brands, this is also a certain pattern that clearly speaks of exactly what and how to sell such a person. Modern recommendation systems aim to detect such patterns.
This is a good, very promising business. Especially now, when every person has an archive of digital data, according to which his consumer portrait can be accurately built in a split second. An archive of digital music in an MP3 player, a collection of digital films, a history of surfing, tags and bookmarks on Web 2.0 sites - and you're done. After analyzing this information, the person fell on the hook of a marketer. Now he will receive commercial offers, which he will not be able to refuse. The point is small - to create a universal system that can automate this process.
Hundreds of different companies are working on this challenge. Even former Kiev resident Max Levchin (founder of Paypal ) created a new startup Slidein this area. He says, by the way, that Paypal is also equipped with something like a recommendation system: they analyze the profile of each user and calculate what the probability of fraud on his part.
Perhaps the next giant of Internet search, the so-called “smart Google” , will be created on the basis of these new wave companies.